Home cleaning KPIs owners should track monthly

What to take away

• A metric without a written definition drifts. Define each one once, in variables, and date any change to the definition.
• Every metric here has to trigger a decision. If nobody would act differently on a bad reading, stop measuring it.
• Six of the ten come from one habit: capturing arrival and departure times on site rather than reconstructing them.
• Watch the trend, not the level. Comparisons to other companies are almost useless because nobody defines these the same way.
• Review five or six monthly rather than twenty annually.

No target, benchmark or industry figure appears here, because none would be true for your market. What the page gives you is the definition and the decision rule.

Define the variables once

Used throughout below, over one month unless stated.

• A for productive on-site crew hours.
• P for crew hours paid.
• T for travel hours.
• J for jobs completed.
• Jo for jobs completed inside their duration allowance.
• Jr for jobs requiring a return visit or touch-up.
• Hr for hours spent on that rework.
• R for revenue collected.
• S for supplies consumed.
• b for burdened cost per crew hour.
• Cn for customers acquired, Cl for customers lost, C0 for customers active at the start.
• M for marketing spend including your own time valued.

The ten metrics

Metric	Formula	Decision it triggers
Utilization	A / P	Falling: tighten route clusters or fix partial days
Allowance accuracy	Jo / J	Falling: the duration allowance or the scope is wrong, not the crew
On-time start rate	On-time first jobs divided by working days	Falling: morning routine or route is over-filled
Rework rate	Hr / A	Rising: a training gap or an allowance too short
Contribution per crew hour	(R - S - b x (A + T)) / (A + T)	Falling: check price, scope and route in that order
Customer retention	1 - (Cl / C0)	Falling: look at cleaner consistency before looking at price
New customers per month	Cn	Flat with strong inquiries: the leak is at response or quote
Cost per acquired customer	M / Cn, split by source	Rising: reallocate between channels using retention, not volume
Employee retention at ninety days	Share of hires still employed at ninety days	Falling: hiring criteria or onboarding, not pay alone
Compliance currency	Share of registrations and certificates in date	Below full: fix immediately, no analysis required

The last row is binary and it is the one people leave off a dashboard. It should sit at the top instead, because every other metric is a question of degree and this one is not. Populate it from your own compliance items with renewal dates register.

The three that tell you the most

Utilization. It is the multiplier on everything. A cleaning company with poor utilization pays the same wages for less sellable output, so its effective cost per productive hour is higher even though its pay rate is identical to a competitor's. It is also the metric most improvable without asking anyone to work harder, because the levers are geographic clustering and morning routine.

Allowance accuracy. This is the bridge between operations and pricing. Persistent overruns mean the allowance was built from an average rather than from the spread, which is a pricing fault that shows up as an operations complaint. Fixing the allowance improves margin, quality and retention at once, because a crew that runs over every day is being told daily that they are failing.

Rework rate. The most honest quality measure in the business, and the one most companies destroy by recording callbacks as separate jobs rather than linking them to the original. Fix the linkage before you trust the number.

Definitions drift, so write them down

Two things go wrong with metrics in small companies.

The first is silent redefinition. Someone starts counting travel differently, and the trend line moves for a reason nobody remembers. Keep a one-page definitions document with the formula, the source record, and a dated note of any change.

The second is measurement changing behavior. Tie pay to a metric and the metric improves faster than the underlying reality does. If you use any of these in compensation, pair it with a counterweight: speed with rework rate, sales with ninety-day retention.

Where the numbers come from

Six of the ten derive from one habit: crews capturing arrival and departure on site at the moment they happen. Reconstructed times are guesses wearing timestamps, and every derived metric inherits the error.

The rest come from ordinary records. Revenue and supplies from the books, kept to the standard set out in the Internal Revenue Service: What kind of records should I keep? guidance. Customer counts from the customer record with an acquisition source captured at inquiry. Employee retention from the people record, alongside training and retention measures. All of it should come out of the systems that produce the KPIs rather than being typed into a spreadsheet from memory.

Guarding the data

The records behind these metrics sit beside home addresses, access codes and a schedule of when houses are empty. Reporting access should be role-limited for the same reason operational access is.

Practical measures at this scale: multi-factor authentication on every account that reaches the data, individual logins, same-day removal when someone leaves, and tested backups. The National Institute of Standards and Technology: NIST Small Business Quick-Start Guides sequence these for small businesses, and the Cybersecurity and Infrastructure Security Agency: Cyber Guidance for Small Businesses material covers what prevents the most common incidents.

The monthly review

Thirty minutes, same day each month, in this order.

1. Compliance currency. Fix anything expired before discussing anything else.
2. Utilization and allowance accuracy together, because they explain each other.
3. Rework rate and customer retention together, for the same reason.
4. Contribution per crew hour by service line.
5. New customers and cost per acquired customer, split by source.
6. Employee retention at ninety days.

Then write one sentence per metric that moved, saying what you will do about it. A review that produces no sentence produced no decision.

Compare each figure to your own last three months rather than to any published average, and check whether the trends still support what you wrote in the plan's assumptions. When they do not, change the plan rather than ignoring the data.

Common questions

How many metrics should a small company track?

Five or six reviewed every month. A dashboard of twenty is a dashboard nobody opens, and the extras rarely change a decision.

Can I compare these to industry benchmarks?

Not usefully. Published figures aggregate companies with different crew models, service mixes and definitions, and definitions are where most of the variation hides. Your own trend is the reliable comparison.

What if I do not have the data yet?

Start with what one habit gives you: on-site time capture. Utilization, allowance accuracy and rework rate all follow from it, and those three cover most of what an owner needs in the first year.

Should crews see the metrics?

Yes, the ones they influence, with the definitions explained. Allowance accuracy and rework rate shown honestly, alongside a willingness to change an allowance that is wrong, builds trust. Shown as a scoreboard with no ability to change the inputs, they build resentment.